Dynamic immune landscape of NAFLD highlights pDCs as predictive biomarkers for disease progression

Non-alcoholic fatty liver disease (NAFLD) is associated with profound alterations in the hepatic immune microenvironment; however, a comprehensive understanding of immune cell dynamics across disease stages remains limited.

We performed an integrative reanalysis of publicly available single-cell transcriptomic datasets from the GEO database, encompassing human liver samples spanning the NAFLD continuum. Candidate immune signatures were validated in independent human cohorts and experimentally induced murine NASH models. A diagnostic classifier leveraging pDC associated gene profiles was developed and rigorously evaluated across multiple validation cohorts.

Analysis revealed extensive stage-dependent remodeling of hepatic immune composition, with pDCs exhibiting the most pronounced transcriptional and numerical shifts. Specifically, a unique pDC subset coexpressing GZMB and TPM2 was significantly enriched in human NASH samples. This signature was consistently recapitulated in independent human datasets and murine NASH models. The pDC-derived diagnostic model demonstrated robust performance for NASH identification across all validation cohorts.

GZMB⁺TPM2⁺ pDCs constitute a conserved, NASH-specific immune signature validated across species. The associated diagnostic model provides a robust, clinically applicable tool for non-invasive NASH identification.